KR960016015B1 - Filter coefficient evaluating method for ghost cancelling - Google Patents

Abstract

averaging received reference signals several times; obtaining the maximum change amount of the filter coefficient by comparing the (t)-th filter coefficient with the (t+1)-th filter coefficient; calculating the threshold value to renew the filter coefficient in the (t+1)-th calculation process by using the maximum change amount and the interface value; setting the filtered coefficient as zero in case that it is smaller than an added threshold value; renewing the filter coefficient value if the change amount of the filter coefficient is larger than the threshold value, and otherwise, replacing it with the former filter coefficient value, in case that the filter coefficient value is larger than the added threshold value; and performing a filtering as to the input signal with the calculated filter coefficient in case that the renewal number is above N.

Description

Calculation method of filter coefficients for ghost removal

1 is a block diagram showing a general ghost removal device.

FIG. 2 is a block diagram showing the detailed configuration of the filter in FIG.

FIG. 3 is a detailed block diagram of the filter coefficient calculation unit in FIG.

4 is a flowchart illustrating a filter coefficient calculation method for ghost removal according to the present invention.

* Explanation of symbols for main parts of the drawings

1: analog / digital converter 2: reference signal extractor

3: filter coefficient calculation unit 4: filter

5: Digital / Analog Converter

The present invention relates to a ghost removal apparatus, and more particularly, to a filter coefficient calculation method for preventing the filter coefficient for ghost removal from fluctuating by white noise and impulsive noise of a channel.

In general, a broadcast signal transmitted from a broadcasting station is reflected by a terrain building such as a building or a mountain while reaching the receiver through the air or a track, or reflected due to the impedance matching of the transmission line not being performed. channel) so that reflected signals other than the original signal appear on the receiver. The phenomenon in which reflected signals other than the original signal appear on such a receiver is called ghost.

Therefore, as part of a method of eliminating ghosting caused by a multipath channel, a broadcasting station carries a ghost elimination reference signal on a part of a transmitted broadcasting signal, and a ghost eliminating apparatus at a receiving side transmits the ghosting signal transmitted from a previously stored broadcasting station. The characteristics of the transmission path are determined by comparing with the ghost removal reference signal identical to the removal reference signal. The filter of the ghost elimination device compensates for the characteristics of the identified transmission path, and when the broadcast signal passes through the filter, it is possible to correct distortion of the broadcast signal generated in the transmission path.

In other words, the ghost elimination and the principle ensure the distortion of the signal generated in the signal transmission process by constructing a filter having the inverse function of the transmission channel transfer function to obtain the transfer function of the transmission channel and to compensate the characteristics of the transmission channel.

1 is a block diagram showing a general ghost removal apparatus. Referring to FIG. 1, the analog / digital converter 1 converts an analog video signal from a broadcast signal received through an antenna into a digital signal and applies the same to the reference signal extractor 2 and the filter 4. The reference signal extracting unit 2 extracts the ghost removing reference signal inserted into a part of the broadcast signal during transmission and applies the extracted signal to the filter coefficient calculating unit 3. The filter coefficient calculating unit 3 compares the extracted ghost elimination reference signal with a prestored ghost elimination reference signal, calculates a filter coefficient for compensating for distortion of a signal generated in a transmission path, and applies the filter coefficient to the filter 4. In the filter 4, the digital video signal output from the analog / digital converter 1 is filtered by the filter coefficient output from the filter coefficient calculator 3 and applied to the digital / analog converter 5. The detailed configuration of the filter 4 will be described with reference to FIG. 2. FIG. 1 shows that the FIR (Finite Impulse Response) filter 41 performs the filter coefficient calculation unit 3 on the signal output from the analog / digital converter 1. Proximity ghost and line ghost are removed and output to the adder 42 by the output first filter coefficient, and an lnfinite impulse response (IIR) filter 43 is a filter coefficient calculator 3 for the signal output from the adder 42. The long ghost is removed using the second filter coefficient output from the N-th filter and applied to the adder 42. The adder 42 adds the output signal of the FIR filter 41 and the output signal of the IIR filter 43 and outputs it to the digital / analog converter 5. The digital / analog converter 5 converts the digital video signal filtered by the filter 4 into an analog video signal and outputs the analog video signal.

The principle of the ghost elimination apparatus will be described with reference to the filter coefficient calculation unit 3 as follows.

Ghosts are usually in the range of -2 to 40 usec in time. In this case, the negative time zone indicates the location of the line ghost. Therefore, in order to remove the ghost in the range of -2 to 40usec, the broadcast station inserts a ghost removal reference signal into a part of 1H (about 20usec), and extracts a signal of 1H period when extracting the received reference signal. When the signal of the 1H period is sampled at a frequency four times the color subcarrier frequency (fsc = 3.58 MHz) of the broadcast signal, 910 data can be extracted. Considering the proper ghost rejection range, the ghost rejection reference signal should be used within 300 samples.

In general, the ghost removal reference signal is transmitted on one line in a section in which no video signal exists. Normally, it should be carried on a specific same line (18H line, 28H line). However, due to various circumstances such as control signal information of the broadcasting station itself or TTX (Teletext) signal, it is difficult for all broadcasting stations to transmit the ghost elimination cancellation signal on a specific same line.

A field calculator (not shown) among the components of the filter coefficient calculator 3 removes the synchronization signal, color burst and noise included in the 1H period of the ghost removal reference signal extracted by the reference signal extractor 2. Inverts the extracted ghost removal reference signal to add it. By adding an inverted ghost cancellation reference signal, the extracted ghost cancellation reference signal remains no ghost generated during transmission without noise. The correlator (not shown) has a maximum correlation value due to the characteristics of the ghost cancellation reference signal when the correlatlon is added to the ghost removal reference signal and the ghost removal possible range of the receiver. For the rest, the correlation value is zero. Therefore, the correlation between the received ghost removal reference signal and the pre-stored ghost removal reference signal is a reference position for ghost removal. The ghost removal reference signal extracted according to this reference position is moved in the opposite direction by the maximum correlation position to match the starting point of the extracted ghost removal reference signal and the previously stored ghost removal reference signal. Methods for obtaining coefficients of the filter 4 using the extracted ghost removal reference signals include a method using an FFT, a Least Mean Square (LMS), and a zero-forcing method. As an example, a method using an FFT will be described with reference to the block diagram of FIG. 3.

Based on the data value of the maximum position obtained by the correlator, the signal output from the field operator is taken as much as the range for removing the near ghost including the line ghost. To remove line ghosts up to -2usec, data is taken before the maximum position of about 30 samples, and afterwards, data is taken from the number of samples occupied by the original ghost removal reference signal. That is, in this case, the range of samples taken to obtain the coefficients of the FIR filter 41 ranges from 30 samples to the range of 20 samples added to the ghost removal reference signal based on the position of the maximum value.

The FFT of this sample data can be converted into data on the frequency axis. The original ghost cancellation reference signal is divided on the frequency axis with the data on the frequency axis. IFFT is performed again, the coefficients of the FIR filter 4l are converted. You can get it. The method of obtaining the coefficients of the IIR filter 43 is first based on the maximum correlation value, starting with the data of the line ghost elimination range (e.g., -30 samples), as in the FIR filter 41, of the ghost elimination reference signal. To remove long ghosts that are about 40usec in length, FFT is performed by taking data in the range of 600 samples plus the ghost removal reference signal. The data on the frequency axis is multiplied by the FFT coefficient of the FIR filter 41, divided by the original ghost elimination reference signal, subtracted from 1, and then clamped to a value higher than or equal to a predetermined level to perform IFFT again. The coefficient of can be found.

As described above, an eight-field calculated reception reference signal, which is a signal input to obtain a filter coefficient for removing a ghost, is input by averaging two or more times to minimize the effect of white noise.

However, the calculated filter coefficient has a problem in that the lower 1 to 2 bits may fluctuate due to noise, thereby degrading reproduction quality.

In addition, since the time required for the calculation of the filter coefficient is relatively long, when impulsive noise occurs at the moment of storing the received reference signal, there is a problem that adversely affects the filtered image quality.

Accordingly, an object of the present invention is to set a threshold value for the variation amount of the filter coefficient calculated by the filter coefficient calculation method such as the sequential correction method or the batch correction method in the ghost elimination device in order to solve the above-described problems, It is to provide a filter coefficient calculation method for minimizing the variation of the filter coefficient due to white noise and impulsive noise by updating only the filter coefficients.

According to an aspect of the present invention, there is provided a filter coefficient calculation method for preventing a filter coefficient from being fluctuated by white noise of a channel, comprising: a first step of averaging a received reference signal by a predetermined number or more; And comparing the t-th filter coefficient and the t + 1 th filter coefficient with respect to the entire i in the t-th calculation process of the specific filter coefficient calculation method using the reference signal averaged in the first step to determine the maximum change amount of the filter coefficient. A third step for calculating a threshold value for updating the filter coefficient in the t + 1 th calculation process using the maximum change amount and the boundary value of the filter coefficient obtained in the second step; A fourth step of comparing a coefficient value with a weighted threshold value that is a value obtained by multiplying the threshold value obtained in the third step and setting the value to zero when the filter coefficient value is smaller than the weighted threshold value; If the filter coefficient value is greater than the weighted threshold value in the fourth step, the filter coefficient value is updated if the variation amount is greater than the threshold value calculated in the third step. If not, the fifth step of replacing the previous filter coefficient value; And a sixth step of performing filtering on the input signal using the calculated filter coefficient when the update frequency of the filter coefficient is greater than or equal to the predetermined number N by the fifth step.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

4 is a flowchart illustrating a filter coefficient calculation method for ghost elimination according to the present invention, in which step 100 is a step for averaging the received reference signal by a predetermined number or more, and step 200 is performed in step 100. In the t-th calculation process of the specific filter coefficient calculation method using the averaged reference signal, a step for calculating the maximum change amount of the filter coefficient is performed. Computing the threshold for updating the coefficients, steps 400 and 500 are for comparing the current filter coefficient value and the weighted threshold value, steps 600 to 800 are for the currently calculated filter coefficients The fluctuation amount is compared with a threshold obtained in step 300. In steps 900 and 1000, the filter coefficient calculated when the filter coefficient update frequency is N or more times is input to the filter. A step for.

The operation of the present invention will now be described with reference to FIG. In the present invention, in order to minimize the fluctuation of the filter coefficient due to white noise, the coefficients calculated by the filter coefficient calculation method such as the sequential correction method or the batch correction method are compared with the previous coefficients.

First, in step 100, the received reference signal is averaged and input for a predetermined number or more.

In step 200, the filter coefficient at the i-th position in the t-th iteration according to a specific filter coefficient calculation method is assumed to be h _t (i). Comparing the t-th filter coefficient and the t + 1 th filter coefficient with respect to the whole i, calculate δ as shown in the following equation (1)

delta = max [mim {2ε _t , | h _t (i) -h _{t + 1} (i) | … … … … … … … … … … (One)

Where ε _t is the boundary value described below. Find the maximum change in filter coefficients through (1). At this time, the change amount larger than 2ε _t was excluded in the process of calculating the maximum change amount on the assumption that the filter coefficient value calculated by the new ghost. In the 300th step, ε _{t + 1} is obtained by using the weighted average of δ and ε _t obtained from _Eq .

ε _{t + 1} = (1-α) ε _t + α δ... … … … … … … … … … … … … … … … … … … … … … (2)

The reason why the weighted average is taken as in _Eq. (2) is to prevent the boundary value ε _t from changing so much. Ε _{t + 1} obtained from _Eq. (2) is used as the threshold for updating the filter coefficient in the t + 1th calculation process.

In operation 400, if the current filter coefficient value is smaller than the weighted threshold value ε _{t + 1} , the current filter coefficient value is compared with the weighted threshold value ε _{t + 1} , and is replaced with zero (step 500). Here, values between 0.7 and 1 are taken as values greater than zero and less than or equal to one. In step 600, the coefficients are calculated by the filter coefficients calculated by the filter coefficient calculation method for ghost removal, and the coefficients are updated when the threshold is greater than the threshold ε _{t + 1} (step 800). Replace with the previous filter coefficient value (step 700). This relationship can be expressed by the following equation (3).

if | h _{t + 1} (i) | <ε _{t + 1} , h _{t + 1} (i) = O;

else if | h _{t + 1} (i) -h _t (i) | <ε _{t + 1} , h _{t + 1} (i) = h _t (i); … … … … … … … … … … … (3)

The filter coefficient calculation method for preventing the variation of the filter coefficient due to impulsive noise is as follows.

Since the received reference signal is input by averaging more than two times, and the filter coefficient calculation time is relatively long, the process of calculating each filter coefficient is calculated by using the received reference signal input at a sufficiently separated time. . Therefore, impulsive noise may be considered to exist in the calculation process of 1 to 2 filter coefficients.

In operation 900, the calculated filter coefficient is filtered (4 in FIG. 1) only when the filter coefficient h _{t + 1} (i) obtained through Equation (3) is updated N or more times in consideration of the characteristics of the impulsive noise. In order to be used for the actual filtering by inputting to (1000). That is, assuming that the filter coefficient update frequency is r _{t + 1} (i), the filter coefficient update expression of Eq.

if | h _{t + 1} (i) | <ε _{t + 1} {h _{t + 1} (i) = 0, r _{t + l} (i) = 0; }

else if | h _{t + 1} (i) -h _t (i) | <ε _{t + 1} {h _{t + 1} (i) = h _t (i), r _{t + 1} (i) = 0; }

else if r _{t + 1} (i) N, r _{t + l} (i) = 0;

else {h _{t + 1} (i) = ht (i); r _{t + 1} (i) = r _{t + 1} (i) +1; }… … … … … … … … … … … … … (4)

Where N is a value between 1 and 2.

As described above, in the filter coefficient calculation method for ghost removal according to the present invention, a threshold value for the variation amount of the filter coefficient calculated by the filter coefficient calculation method such as the sequential correction method or the batch correction method is set to generate a filter having a variation amount greater than or equal to the threshold value. By updating only the coefficients, it is possible to minimize the change of the filter coefficient due to the white noise, and to input only the filter coefficient whose number of times exceeding the threshold is more than N times to the filter to minimize the change of the filter coefficient due to the impulsive noise. There is an advantage to this. Here, the threshold may be adaptively set to be variable according to the characteristics of the channel.

Claims (8)

CLAIMS 1. A filter coefficient calculation method for preventing a filter coefficient from being changed by white noise and impulsive noise of a channel, comprising: a first step of averaging a received reference signal by a predetermined number or more; In the t-th calculation process of the specific filter coefficient calculation method using the reference signal averaged in the first step, a maximum change amount of the filter coefficient is obtained by comparing the t-th filter coefficient and the t + 1 th filter coefficient with respect to the entire i. A second step for; A third step of calculating a threshold value for updating the filter coefficient in the t + 1 th calculation process using the maximum change amount and the boundary value of the filter coefficient obtained in the second step; A fourth step of comparing the currently calculated filter coefficient value with a weighted threshold value which is a value obtained by multiplying the threshold value obtained in the third step and setting the value to zero when the filter coefficient value is smaller than the weighted threshold value; If the filter coefficient value is greater than the weighted threshold value in the fourth step, the filter coefficient value is updated if the variation amount is greater than the threshold value calculated in the third step. Otherwise, the fifth step for replacing the previous filter coefficient value, and the sixth step for performing filtering on the input signal with the filter coefficient calculated when the update frequency of the filter coefficient is greater than the predetermined number N by the fifth step. A filter coefficient calculation method, characterized in that it comprises a step. The method of claim 1, wherein the maximum change amount δ of the filter coefficient is δ = max [mim {2ε when the filter coefficient at the i-th position is h _t (i) and the boundary value is ε _t in the t-th calculation process. _t , | h _t (i) -h _{t + 1} (i) |}. 3. The method of claim 2, wherein the threshold value is obtained by a weighted average of the maximum change amount of the filter coefficient and the boundary value. 4. The filter coefficient calculation method according to claim 3, wherein in calculating the threshold value, a variation of more than a specific value is excluded. The method of claim 4, wherein the specific value is 2ε _t . 6. The method of claim 5, wherein the threshold value is variably set according to the characteristics of a channel. 2. The method of claim 1, wherein the weight is set to a value between 0.7 and 1 with a value greater than zero and less than or equal to one. The method of claim 1, wherein the predetermined number of times (N) is set to 1 to 2.